Hi all, It turns out that this is because the map projection scripts call the curvilinear SHADE command with the mapping variables computed based on the coordinates, and what we'd prefer here is to have the coordinates computed using the cell edges. Here's the bit from the Users Guide (see SHADE, curvilinear version in the index) ...
2. In the 3-argument SHADE syntax you can specify either the
coordinates yes? SHADE values, xcoords, ycoords
say that nVx is the size of the "x" dimension of the values
argument, and
If nCx = nVx then the xcoords argument is presumed to give the
locations However, if nCx = nVx + 1 then the xcoords and
ycoords arguments are the edges instead. This would be easy if we had a pseudo-variable with the x and y edges - something that's on our list of things to do but which we haven't yet got. Muyin Wang wrote: Hello, I have been trying to plot shaded plots over polar stereographic projection using go mp_stereographic_north 0 90; set grid myvar shade myvar,x_page,y_page go mp_graticule 0 360 90 50 90 10 It seems that ferret always plot the outside grid box in half no matter how do I define the region, regardless the underlying grid box. Attached 2 figures are based on CRU SAT data (5x5 degree resolution), and the first one was created by def region/y=47.5:90, while the 2nd one was created by set region/y=50:90n I also tried using NCEP Reanalysis data for a similar test. Unfortunately, ferret does the same thing: the most outside of the domains are shaded by half of the grid box (NCEP's grid box is 2.5x2.5). How could I get the full grid shaded in this map projection? Thanks |